nRF9151 DECT NR+ PHY: dect rssi_scan reads a CW tone ~40 dB below the conducted input level

[DVH]

Setup

• Two nRF9151 SMA DKs (LACA A1), SDK v3.4.1.
• Board B: mfw-nr+-phy_nrf91x1_2.0.0, running the DECT NR+ PHY Shell sample (samples/dect/dect_phy/dect_shell, unmodified).
• Board A: LTE mfw_nrf91x1_2.0.4, used only as a CW source via AT%XRFTEST TX test, band 25, around 1890.4 MHz (inside NR+ band 1, channel 1667).
• Conducted path: A port → 3 dB → 10 dB or 30 dB → 3 dB → B port, plus short cables.
• Every attenuator and cable was checked individually with a tracking generator and spectrum analyzer (each within 1.5 dB of nominal).
• Board A’s CW output, measured on the spectrum analyzer through the same pads: −9.1 dBm at a −10 dBm setting.
• Board B’s NR+ TX output (dect rf_tool -m tx, measured as channel power on the analyzer): within ~2 dB of its setting.

Measurement: dect rssi_scan -c 1667 -t 1000 on board B.

CW level at B’s port (dBm)	RSSI highest / lowest (dBm)
~−75	−104 / −109 (noise floor)
~−65	−101 / −106
~−55	−95 / −96
~−45	−86
~−25	−66

What we’ve ruled out

• Linearity and repeatability: the reading tracks the input 1 dB per dB over about 40 dB. Swapping the 30 dB pad for 10 dB moved every reading by exactly 20 dB. It repeats within ±1 dB.
• Tone position: placing the tone anywhere from −432 to +368 kHz from the channel centre makes no difference.
• Leakage and saturation: opening the link gives noise only, and no saturations are reported.

Questions

1. How is the RSSI in dect rssi_scan defined? Is it total power in the 1.728 MHz channel, per subcarrier, or something else? Is it referenced to the antenna port?
2. Is a ~40 dB under-read expected for a narrowband (CW) signal? Does a real NR+ (OFDM) signal read correctly?
3. Does the expected-RSSI / RX gain setting affect the reported value?

[DVH]

PS: After significant extra measurements, we have a sharper view of the situation:

The offset depends on frequency. Band 4 reads correctly; band 1 is ~40 dB low.

I repeated the test on NR+ band 4. Same two DKs, board B unchanged except dect sett -b 4 followed by deactivate/activate. Board A sent a CW tone at 912.4 MHz (LTE band 8 TX test), and B ran dect rssi_scan -c 535 (912.384 MHz). The path was 3 dB + 10 dB + 3 dB plus cable, about 16.5 dB.

CW level at B’s port	rssi_scan result
~−55.5 dBm	−56 / −57 dBm (correct within ~1 dB)
~−45.5 dBm and above	saturated (high −127, low +1, saturation counts reported)

Summary across both tests

• Band 4 (912 MHz): the CW reading is correct. Saturation starts somewhere between −55 and −45 dBm at the port.
• Band 1 (1890 MHz): the CW reading is ~40 dB low. There is no saturation even at about −25 dBm at the port, so the signal appears genuinely ~40 dB weaker inside the receiver, not just reported low.
• Board B’s transmit at 1890 MHz (dect rf_tool -m tx, measured on a spectrum analyzer) is within ~2 dB of its setting, so the shared SMA path and the transmitter are fine.
• The tone position within the channel makes no difference at 1890 MHz.

So the problem appears to be receive-only, at 1.9 GHz, on an nRF9151 SMA DK with mfw-nr+-phy_nrf91x1_2.0.0 and the unmodified PHY Shell sample (SDK v3.4.1).

Questions

1. Is there a known receive-path or front-end configuration issue for band 1 on the nRF9151 SMA DK with the NR+ PHY firmware? For example, MAGPIO or antenna-switch setup. The sample sets CONFIG_MODEM_ANTENNA_AT_COEX0="" because the NR+ firmware doesn’t support it. Could the DK’s RF switch be left in the wrong state for 1.9 GHz receive?
2. Is band 1 RSSI expected to need a calibration or RX-gain setting that band 4 doesn’t?

[Simon D-M]

Hi,

It is very strange...

Just to confirm, your setup consists of 2x nRF9151 SMA DK connected directly together via a coax cable? Is there anything more in between? Is it performed in a radio chamber?

Does it also happen if you swap the firmware on the DevKits ?

What RSSI do you measure if you send a real NR+ signal with dect rf_tool -m tx ?

Best regards,

Simon